With successful stationary turns we are taking a look at moving turns this week. The first algorithm had been to run the tires on one side at 100% and the tires on the the other side at 50%.

In testing, this didn’t provide us with any noticeable turn.  The inertia present in the moving motor meant that we were not seeing enough decrease in speed to affect the forward trajectory.

In our next test, we reduced the motors on the inside of the turn to a 0% duty cycle so that they were completely undriven.  This would hopefully cause the turning effect we were looking for.

Test #8 – Moving Turns

Here is the test setup:

Outside Motors (1 & 3):

• Motor 1 = 100%
• Motor 3 = 100%

Inside Motors (2 & 4)

• Motor 2 = 0%
• Motor 4 = 0%

After setting the jumper…

Again, the inertia present in the moving motor was so high that we didn’t get any significant turning action.  The next step was to apply some braking to the inside wheels by reversing the polarity of the motor and trying signals with differing power levels (duty cycles).

Also, for ease of use we are going to replace the jumper normally used to program the Rabbit with a switch sometime soon.  That means we won’t need to grab a set of needle-nose pliers whenever we want to perform a floor test like this one.

Test #9 – Moving Turns with Inside Braking

Here is the test setup:

Outside Motors (1 & 3):

• Motor 1 = 100%
• Motor 3 = 100%

Inside Motors (2 & 4)

• Motor 2 = -50%
• Motor 4 = -50%

This isn’t what we want either.  What we get here is forward motion interrupted by a stationary turn.

What is Next?

We will continue to reduce the inverted power to the inside wheels to get a nice moving turn.  Hopefully, we can find the ‘butter-zone’ somwhere between 0% (freewheeling) and -50% (shown above) that will provide is with a nice moving turn with a quantifiable turning radius.

Yesterday evening we couldn’t get the rover into a successful turn. I went home and thought about the issue but after posting it this morning my friend Allen had a suggestion.

“Have you tried a biscuit treat? It works for my dogs real well.”

Sadly, the dog biscuit didn’t elicit a response the response I was hoping for.

My suspicion was that we weren’t getting enough torque to overcome the lateral friction on the tires.  At 50% power (duty cycle), the pulse we were sending just weren’t big enough.   This morning we cranked the duty cycle up to 99% and this is what we got.

Here is what we tried:

• Turning Test 1: Duty Cycle = 50%  –> shivering sad puppy behavior
• Turning Test 2: Duty Cycle = 50% with additional weight on the rover –> shivering sad puppy behavior
• Turning Test 3: Duty Cycle = 99% –> Successful turn
• Turning Test 4: Duty Cycle = 95% –> Successful turn
• Turning Test 5: Duty Cycle = 90% –> Successful turn (slow)
• Turning Test 6: Duty Cycle = 80% –> Failure (partial turn and stop)
• Turning Test 7: Duty Cycle = 85% –> Failure (partial turn and stop)

Success

That tells us that when we want to execute stationary turns we should probably kick the duty cycle up to 90% or higher.  Now that we have good turning behavior we can move on to quantifying the turns.  We want to be able to spin the device with some degree of precision.